[evlatests] WIDAR-0: amp wobble tracks slot number, NOT antenna/baseline

[Michael Rupen]

Baselines 18-19 and 19-28 are consistently among the worst baselines from
WIDAR-0: very short (10msec) integrations show some 10s of per cent
repeating amplitude wobbles with a frequency of a few Hz in all subbands,
while 1 second dumps show what looks like an aliased version of this, with
somewhat smaller amplitude (perhaps +/- 5%).  The VLA correlator shows no
such effect, even with 0.488 second dumps.

These three baselines are "special" in two ways: they are among the shortest
baselines available with WIDAR-0; and they are the two baselines which
involve the three Station Boards in slot 0 of the three currently-populated
Station Racks.

As reported previously, Kerry swapped the fibers from Station Rack 2, Slots
0 and 1: antennas 28 and 25.  The amplitude wobble stays with the slot
number rather than the antenna -- baselines 18-19 and 19-*25* are now bad,
while 19-28 is now fine.

So we now have two effects which know about the locations of the station
boards: the phase jump, and the amplitude wobble.

Any thoughts on why this should be?

                 Michael

p.s. For WIDAR-0 afficianados, we have not yet checked to see whether the
   subband 0 wobbles also track station board location.

p.p.s. Note that neither the phase jumps nor the amplitude wobbles
   explain the "noise floor" we're seeing in the hour-long images.  That must
   either be a more subtle manifestation of a similar problem, or a
   completely independent difficulty.